[Conversion Plugin] KFX Input

[DNSB]

Please note the last sentence of what I wrote. It's entirely possible that those artifacts have survived Amazon's conversion to KFX and then back to ePub. If you think those are rather stupid, you should take a look at many commercially generate ePubs and their stylesheets—many of which I believe were generated by interns who had never created an ebook before. OTOH, as I have argued several times, how many people who are not professional or serious amateur ebook creators would ever look at the contents of a stylesheet or .opf file?

[jhowell]

Originally Posted by Jaws

I'm encountering three different kinds "unit inconsistencies" in KFX --> ePub; they may be unrelated, they may be due to KFX itself, to the KFX Input plugin, or something in Calibre's own conversion process. Or, I suppose, in something else entirely.

These inconsistencies will likely have multiple causes: the formatting supplied by the publisher, Amazon's conversion of that to KFX, this plugin's conversion of KFX to EPUB, and calibre manipulation of that EPUB. There is no effort made to maintain the sort of consistency that you would like in this process.

Originally Posted by Jaws

1. "Line height" is inconsistently "unitless" in heavily-illustrated books (and this includes, sometimes, tables of data that were embedded as a graphic). In the stylesheet (usually stylesheet.css), some styles will contain each of the following (all of which are/should be valid ePub characteristics):
line-height: 1.29017;
line-height: 129.017%;
line-height: 1.29017em;
This bugs me; it's inconsistent with good laboratory practice — and ability to maintain files over time. It has bugged me about the graphic-design community for several decades, so it may have nothing whatsoever to do with Calibre or any plugin or even KFX.

The reason that I think this may be Calibre is that the first one (the purely unitless one) only appears in "line-height" and margin/padding, never in font size or text-indent. This is a "I'm trying to understand" thing more than a bug report; but...

You can compare various versions of the book if you want to see what changes were introduced by each step. However that could be a lot of work.

The "From KFX" toolbar action can be used to bypass calibre's part of the conversion process to see the CSS produced by this plugin. That will not match exactly the content of KFX format since KFX actually uses something similar to but not the same as HTML/CSS.

To see the actual KFX content in human readable form use the --unpack option of the command line interface supplied by this plugin. "Creating a symbol catalog file" in the first post of this thread has steps needed before this can be accomplished. Making sense of the result is not easy since KFX is undocumented.

You can come close to seeing what the publisher provided to Amazon by obtaining the book in KF8/AZW3 format, available using some older kindle models and apps, and then converting that to EPUB using the KindleUnpack plugin.

Originally Posted by Jaws

2. Conversion to different kinds of margin (and padding) shorthands in the ePub stylesheet seems completely random, even within the same book. One might find all of the following in the same ePub stylesheet (I just did this morning):
margin: 0 0 6.24399%;
margin-left: 6.24399%;
margin-left: 0.5em;
margin: 0 0 0 6.24399%;

This one is closer to a bug report than item 1, but the code still "works" so maybe it's not a true "bug." It is in all probability related, somehow, to "what font was specified in the KFX file, whether that font was embedded, and whether that embedded font was subsetted (possibly who subset it, too)." But it's puzzling; specifically, what's puzzling is which form for {editorial comment: STOOOPID choices in the CSS specification for "shorthands", so definitely not a Calibre issue} stating the "margin" is chosen.

This plugin will not produce a margin property with multiple values since there is no KFX equivalent to that. That combining of individual margin-xxx properties must be due to calibre.

Originally Posted by Jaws

3. Where, oh where, does the "this book sometimes specifies percentages and sometimes specifies ems" business come from? Is there a way to force conversions to choose one or the other?

Those are two different things that have no equivalent conversion. Percent is used for dimensions that should scale with the size of the containing block and ultimately the size of the screen. em is used when the dimension is intended to scale with the font size chosen by the reader. Both are needed for different cases.

[Argel]

With the latest DeDRM and KFX plugins I can import from my Oasis but not, apparently, from a recent Scribe. Is that par for the course?

Argel

[miss_p1nky]

Sorry if it's a bit out of topic but is it safe to update to K4PC 2.7.1?

[jhowell]

Originally Posted by Argel

With the latest DeDRM and KFX plugins I can import from my Oasis but not, apparently, from a recent Scribe. Is that par for the course?

No, it's not. It is possible to import most Kindle ebooks from a Scribe by using the Device view in calibre as long as you have everything set up properly. For the details you should see the Dealing with Kindle ebooks in calibre thread. That is also a good place to ask questions if you are stuck.

[jhowell]

Originally Posted by miss_p1nky

Sorry if it's a bit out of topic but is it safe to update to K4PC 2.7.1?

No it is not. Any Kindle for PC version newer than 2.4.0 uses new DRM for which no solution is currently available.

[miss_p1nky]

Originally Posted by jhowell

No it is not. Any Kindle for PC version newer than 2.4.0 uses new DRM for which no solution is currently available.

Got it, thank you!

[JSWolf]

Originally Posted by Jaws

In case it wasn't clear, I understand that these are all different "acceptable" descriptions for "the same thing" in CSS. (I'll growl some other time about the internal logic of CSS shorthands.) My confusion is at why these are showing up inconsistently in the same document; I do not think it's unique to KFX --> ePub, just more common and more noticeable — and more prone to break during later file maintenance/updating, let alone future formats and devices. (Not even 1992-vintage Quark was this internally inconsistent with measurements.)

I don't care for margin shortcuts. I prefer it typed out the long way. But I sometimes I will edit it and change it from the shorthand.

[DNSB]

Originally Posted by Argel

With the latest DeDRM and KFX plugins I can import from my Oasis but not, apparently, from a recent Scribe. Is that par for the course?

Do you have both the Oasis and Scribe serial numbers in your DeDRM configuration?

[Argel]

When attempting to convert certain KFX formatted books - not all - I'm getting errors of the kind shown below. Latest KFX plugging installed.
Argel

calibre, version 8.1.1
ERROR: KFX conversion failed: <p><b>[book name redacted]</b><br><br>Exception('box-align right with float right in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align left with float left in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /c60.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align left with float left in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml\nbox-align right with float right in /cWH.xhtml')</p>

[jhowell]

Originally Posted by Argel

When attempting to convert certain KFX formatted books - not all - I'm getting errors of the kind shown below. Latest KFX plugging installed.
Argel

calibre, version 8.1.1
ERROR: KFX conversion failed: <p><b>[book name redacted]...

I have not seen that error. I would help to know the ASIN of one or more books exhibiting this problem.

[Argel]

Originally Posted by jhowell

I have not seen that error. I would help to know the ASIN of one or more books exhibiting this problem.

This?

mobi-asin:B002RI9HR0

Argel

[Argel]

calibre, version 8.1.1
ERROR: KFX conversion failed: <p><b>Cannot convert The State We're In</b><br><br>Exception("section section-0_1_e06d48a9d308d92c_3055, $269 9178, story story-0_1_d7238a2033f57c48_3054, $269 7332, $276 7475 content type $276 has extra data: ($102: ($306: $308, $307: Decimal('0')))")</p>

mobi-asin:B00ABLJ2GG

[jhowell]

Originally Posted by Argel

B002RI9HR0

Originally Posted by Argel

B00ABLJ2GG

Unfortunately, neither of those is available from Amazon in my country (USA).

[jhowell]

Version 2.23.0 - 11 Apr 2025

Handle extra formatting that occurs with images in some books that were converted to KFX by Amazon in 2015. Fixes "Unexpected image style for SVG wrapper".

Handle list-indent when present in a list element. Fixes "content type $276 has extra data: ($102: ($306: $308, $307: Decimal('0')))".

Fix error reporting sometimes failing when using "From KFX". Fixes "AttributeError: 'JobLog' object has no attribute 'html'".

Do not fail to convert books that contain invalid control characters.

Omit unexpected empty entries from the list of fonts in a font-family property.